a program to find, within an alignment of large sequences, those regions with an informative content similar to the displayed by the whole alignment and select, among them, the shortest empiricaly significant ones